In that case you are at an impasse, I'd say.  ext3 doesn't work with 
Squid (it may have improved in later versions...try the Red Hat 2.4.18 
kernel--I was able to get it built and installed on a 7.2 system with 
only two package upgrades to satisfy dependencies).
ReiserFS is trustworthy, in my experience.  I have about 40 Squid boxes 
in the field, some for two years, all running on ReiserFS.  I have only 
seen one FS corruption...and it was due to a faulty disk that had to be 
replaced.  In fact, given the troubles I've had with ext3, ReiserFS is 
the only filesystem I trust for Squid.  (I have a little test box 
locally that started reporting disk full when there is 10GB of free 
space, causing Squid to exit...it's running ext3).
But then, as I said, it is wise to recompile with ReiserFS built into 
the kernel, rather than as a module.  I always have, because there was a 
time (maybe even as late as 2.4.9) that ReiserFS would cause problems 
when built as a module.  Turn off the checks and proc interface, as well.
Just MHO, of course.
Andreas Jung wrote:
> And I don't trust reiserfs...I have seen two boxes with corrupted
> reiserfs partitions...
> 
> Andreas
> ----- Original Message -----
> From: "Joe Cooper" <joe@swelltech.com>
> To: "Andreas Jung" <andreas@zope.com>
> Cc: <squid-users@squid-cache.org>
> Sent: Tuesday, May 21, 2002 13:54
> Subject: Re: [squid-users] High load on squid machine after switching from
> reiserfs to ext3
> 
> 
> 
>>Yep.  Don't use ext3.  ;-)
>>
>>I've been able to crash kernel 2.4.9 (the Red Hat RPM kernels) at will
>>when running a Squid workload on ext3.  I never looked very deeply into
>>it, because the performance of ext3 isn't comparable to ReiserFS.  It
>>may be worth your time to grab the latest kernel SRPM, and rebuild it
>>with ReiserFS built into the kernel (rather than as a module).
>>
>>I've had no problems with 2.4.9-31, on a quite large number of boxes.
>>
>>Andreas Jung wrote:
>>
>>>We are running Zope behind Squid 2.4Stable6 with squid in acceleration
>>
> mode.
> 
>>>The squid box (dual Pentium III 1 GHz, RH 7.2, Linux 2.4.9-21smp, 2GB
>>
> Ram)
> 
>>>has during busy hours a normal load of 0.2-0.3 . From time to time
>>>we see spikes over some hours where the load average of the machine
>>>is higher than 1.5 although there are no spikes in the CPU
>>>utilization. Also there is no increase in the number of incoming
>>
> requests.
> 
>>>The only change we made was the switch from reiserfs to ext3 for the
>>
> cache
> 
>>>partition since reiserfs corrupted its partition some time ago.
>>>
>>>Any insights on this issue?
>>>
>>>Thanks,
>>>Andreas
>>
>>--
>>Joe Cooper <joe@swelltech.com>
>>Web caching appliances and support.
>>http://www.swelltech.com
>>
>>
> 
> 
-- Joe Cooper <joe@swelltech.com> Web caching appliances and support. http://www.swelltech.comReceived on Tue May 21 2002 - 12:25:17 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:08:11 MST